Somnologie 10: 36–42, 2006

EEG Slow Wave Activity Regulation in Major Depression EEG-Slow-Wave-Aktivita¨t bei Patienten mit Major Depression Christoph Nissen1,2, Bernd Feige1, Eric A. Nofzinger2, Ulrich Voderholzer1, Mathias Berger1, and Dieter Riemann1 1

Department of Psychiatry and Psychotherapy, Albert-Ludwigs-University Freiburg, Germany Western Psychiatric Institute and Clinic, University of Pittsburgh School of Medicine, USA

2

Summary

Question of the study Sleep disturbances, including reduced slow wave activity (SWA), are among the most consistently replicated biological alterations in major depression. According to the two-process model of sleep regulation, SWA is homoeostatically regulated as a function of prior waking time. This study aimed to further elucidate SWA regulation in major depression. Methods All-night spectral analysis was performed before and after therapeutic sleep deprivation in N ¼ 20 patients suffering from major depression. Recovery sleep was studied at 1700 h after 34 h of waking time (n ¼ 10) and at 0200 h after 43 h of waking time (n ¼ 10). Results A significantly lower increase of SWA from baseline to recovery sleep was observed after 43 h versus 34 h of waking time. Conclusion The finding of a lower SWA increase after 43 h versus 34 h of waking time may indicate a circadian modulation of SWA and is in line with previous findings of studies investigating extended sleep durations or the relationships of SWA and circadian endocrine systems. Keywords depression – sleep – EEG – slow wave activity – circadian rhythm – twoprocess model

Zusammenfassung

Fragestellung Schlafsto¨rungen, einschließlich reduzierter EEG-Slow-Wave-Aktivita¨t (SWA), geho¨ren zu den am besten replizierten biologischen Auffa¨lligkeiten depressiver Sto¨rungen. Gema¨ß dem Zwei-Prozess-Modell der Schlafregulation unterliegt die SWA einer homo¨ostatischen Regulation in Abha¨ngigkeit der vorangegangenen Wachzeit. Ziel der Studie war es, die Regulation der SWA bei Patienten mit Major Depression na¨her zu untersuchen. Methoden Eine Spektralanalyse des Schlaf-EEGs wurde bei N ¼ 20 depressiven Patienten vor und nach therapeutischem Schlafentzug durchgefu¨hrt. Der Erholungsschlaf wurde zu zwei unterschiedlichen Zeitpunkten aufgezeichnet, um 17.00 Uhr nach 34-h-Wachzeit (n ¼ 10) und um 2.00 Uhr nach 43-h-Wachzeit (n ¼ 10). Ergebnisse Nach 43-h-Wachzeit wurde ein signifikant geringerer Anstieg der SWA gemessen als nach 34-h-Wachzeit. Diskussion Die Beobachtung eines geringeren SWA-Anstiegs nach la¨ngerer Wachzeit ko¨nnte auf eine zirkadiane Modulation der SWA hinweisen. Schlu¨sselwo¨rter Prozess-Modell

Depression – Schlaf – EEG – Slow-Wave-Aktivita¨t – zirkadian – Zwei-

Correspondence: Christoph Nissen, MD, Western Psychiatric Institute and Clinic, 3811 O’Hara Street, Pittsburgh, PA 15213-2593, USA Tel.: +1-412-586 9844, Fax: +1-412-246 5300, E-mail: [email protected] Received: 27.07.05/Accepted: 08.12.05


2006 Blackwell Verlag, Berlin

www.blackwell-synergy.com

EEG Slow Wave Activity Regulation in Depression

Introduction Characteristic alterations of sleep in depressed patients, mainly the enhancement of rapid eye movement (REM) sleep and an attenuation of slow wave sleep measures (for overview, see [6]), have markedly contributed to the conceptualisation and pathophysiological understanding of affective disorders. For these studies, the exact knowledge of fundamental processes of sleep regulation represents the basis for further insights into pathophysiological processes of affective disorders. Among various sleep parameters, the electroencephalographic activity in the frequency range from 1 to 4.5 Hz (slow wave activity, SWA) of non-REM (NREM) sleep is of special interest for three reasons: First, the SWA of NREM sleep was found to be reduced in depressed versus healthy subjects (cf. [3, 24]). Second, depressed patients demonstrated an increase of SWA from the first to the second NREM sleep episode, whereas a decrease from the first to the second episode is usually found in healthy controls [23, 24]. Third, the distribution of SWA across the first two NREM sleep episodes, expressed as the SWA ratio episode 1 to 2 (delta sleep ratio), has been shown to be of predictive value for the clinical outcome of pharmacotherapy [14], psychotherapy [22, 35], and therapeutic sleep deprivation [26]. In these studies, a higher delta sleep ratio was associated with a more favourable therapeutic outcome, i.e. longer clinical remission or better therapeutic response. However, the causality of the relationship between the reported SWA abnormalities and the polysomnographically registered duration of sleep episodes has to be further evaluated. The most important model of SWA regulation is the twoprocess model of sleep regulation proposed by A. Borbe´ly and colleagues [7]. According to this model, timing and structure of sleep are determined by the interaction of two processes: a sleep-dependent Process S and a circadian Process C. The time course of Process S was derived from SWA measured at sleep onset following different time spans of waking. During waking time, Process S reflects the sleep–wake-dependent part of sleep propensity, which increases as a function of prior waking time. During sleep, Process S represents the sleep intensity with a maximum at sleep onset and an exponential decline during NREM sleep. In contrast to the homoeostatic character of S, Process C represents the sleep–wake-independent, circadian part of sleep propensity with a minimum at 1600 h and a maximum at 0400 h in parallel to the circadian REM sleep propensity. Process C is closely coupled to the body temperature periodicity [8] and mainly regulated by circadian pacemaker cells in the suprachiasmatic nucleus [34, 36]. The two-process model provides in its fundamental tenets a broadly accepted and often replicated description of global sleep patterns and their circadian modulation. However, fitting various electroencephalographic data sets of sleep deprivation studies, Putilov [29] evaluated that the empirical time course of SWA would be slightly better described using a circadian modulation and postulated that the homoeostatic process S is affected by the circadian pacemaker. In the following years, several studies sustained this hypothesis: De Koninck and co-workers [9] described a significant increase of SWA in the final 3 h of sleep using an extended 15-h sleep protocol in healthy subjects. The SWA return in late sleep was not related to preceding amounts of waking time after sleep onset or REM sleep but was associated with the circadian modulation of the core body temperature. Furthermore, an association of SWA and circadian hormone systems could be demonstrated in form of a positive correlation Somnologie 10: 36–42, 2006

37

between SWA and the growth hormone (GH) release (for GH-releasing hormone in the rat, see [28]; for GH in human subjects, see [19]) and the growth hormone cortisol secretion ratio [32] (for review, see [33]) and a negative correlation between SWA and melatonin secretion [12]. In order to contribute to a better understanding of SWA regulation and as part of a larger study on the clinical stabilisation of SD response [31], we investigated the sleep EEG of depressed patients before and after therapeutic sleep deprivation (SD) by spectral analysis. For the present analysis, SD non-responders were excluded to prevent biasing due to different psychopathology. SD responders underwent a sleep phase advance (SPA; recovery night-time in bed scheduled from 1700 to 2400 h after 34 h waking time) or a sleep phase delay protocol (SPD; recovery night-time in bed from 0200 to 0900 h after 43 h waking time). According to the Process S hypothesis, we expected the increase of SWA to be lower in the SPA than in the SPD group.

Subjects and methods Patient sample Twenty hospitalised patients with a major depressive disorder without acute suicidal or psychotic features according to DSM-IV-criteria [2] were studied. The analysis was part of a study on the stabilisation of the antidepressant effect of sleep deprivation by shifting the sleep phase (advance versus delay) after SD response (clinical findings published elsewhere; see [31]). Bipolar patients, patients with other psychiatric disorders, sleep disturbances, or substance abuse were excluded. All patients included were free of any psychoactive medication for at least 7 days prior to the beginning of the study (in the case of fluoxetine and neuroleptics, 21 days). Before entering the protocol, all participants underwent an extensive physical examination, including an electrocardiogram (ECG), electroencephalogram (EEG), and laboratory screening to rule out any relevant somatic disorder. In addition, a urine drug screening demonstrated that all participants were free of any benzodiazepines, barbiturates, amphetamines, or opiates. Only women with a negative pregnancy test and sufficient contraception were included. Of the 57 patients who took part in a polysomnographic study with at least an adaptation and a baseline night in the sleep laboratory and a trial of therapeutic sleep deprivation, n ¼ 20 were retained for the present spectral data analysis. This reduction of n was due to the following methodical criteria: limitation to a single type of signal recording hardware, exclusion of recordings with unstable electrode impedances in the C3 lead, limitation to SD responders (n ¼ 40), strict SD response criteria, and precise matching (see below). As the sleep EEG spectral power has been shown to depend on age and sex [4, 15, 25, 27], we carefully pairmatched the patient groups (advance versus delay) with respect to these parameters. The resulting close correspondence of demographic and clinical characteristics of our samples is shown in table 1. All patients were informed in detail about the experimental procedures and gave their informed written consent. The study had been approved by the local ethics committee. Experimental design All patients slept 2 nights in the sleep laboratory prior to SD. Before the first sleep laboratory night, habitual bed times

38

Christoph Nissen et al.

Table 1. Demographic and clinical characteristics of investigated samples (mean ± SD)1. Sleep phase advance N Male/female (n)

10 5/5

Sleep phase delay 10 5/5

Significance of the difference1 – –

Age (years)

42.9 ± 7.3

39.7 ± 10.8

0.448

Age at first depressive episode (years)

36.0 ± 9.1

34.2 ± 9.9

0.677

Duration of affective illness (years)

6.9 ± 10.3

5.5 ± 5.4

0.707

Duration of acute episode (weeks)

9.0 ± 5.1

10.5 ± 8.3

0.633

Severity of acute episode (21-Hamilton at baseline)

28.0 ± 4.1

24.6 ± 4.6

0.101

6-Hamilton before SD (average 0900/1600 hrs)

9.0 ± 3.2

8.2 ± 2.5

0.537

6-Hamilton after SD (average 0900/1600 h)

1.8 ± 1.0

2.8 ± 1.6

0.130

1

t-test for independent samples

were ensured by the inpatient conditions (about 2300 to 0600 h). During the study, bed times were strictly controlled under sleep laboratory conditions. After an adaptation night, a baseline night was recorded (2300 to 0600 h), followed by 1 night of total sleep deprivation. Patients were then distributed to a sleep phase advance (SPA; time in bed scheduled from 1700 to 2400 h) or a sleep phase delay (SPD; time in bed from 0200 to 0900 h) recovery night. SPA resulted in a waking time of 34 h, SPD in a waking time of 43 h. Polysomnographic and spectral analytic data are presented for the baseline and recovery nights. To measure psychopathology, the 21-item HAMD was administered prior to the first investigation in the sleep laboratory and repeated the day before SD. Patients were required to have a score ‡18 for inclusion. The 6-item HAMD, an abbreviated version of the 21-item HAMD, is a suitable instrument for repeated measurements [5] and was used to measure depressive mood daily throughout the study at 0900 h and 1600 h. For the present analysis, patients showing at least a 50 % improvement in the 6-HAMD values (averaged 0900 h/1600 h values) comparing the day after to before SD were defined as SD responders; patients showing an improvement